Consanguinity, land inheritance and risk of grandchildlessness

Abstract: Investigating the reproductive consequences of consanguinity in historical Krummhörn, we show social-group specific associations with correlates for fertility and breeding success like the risk of permanent grandchildlessness. While the proportion of mothers, for which not any birth of a grandchild has been documented at all is much lower among socioeconomically privileged farmers, they will even have more often have at least one grandchild born in this population if married consanguineously. However, opposite effects are observed in families without known landholdings for which consanguinity is associated with increased risk of grandchildlessness. We interpret our findings in regard to local resource competition, but also discuss limitations of our study due to data constraints.

Sample selection: Marriages between 1720 and 1830

Starting with the sample from https://gitlab.com/johannes.johow/consanguinity we find the extinction risk (i.e. the proportion of couples not having ever any grandchild born in the study population) peaks for couples married after 1830 (Fig. 1). In the following, only couples married before 1830 have been included in analysis. Note also, that - on average - extinction risk is higher and number of grandchildren ever born is lower among families without known landholdings (‘unknown/landless’) compared to landholding families (‘landowner’).

Average number of grandchildren ever born for 9359 mothers married between 1720 and 1874  by marriage decade and status of landholding (left) and corresponding Probability of grandchildness within the local population.

Figure 1: Average number of grandchildren ever born for 9359 mothers married between 1720 and 1874 by marriage decade and status of landholding (left) and corresponding Probability of grandchildness within the local population.

Total fertility and breeding success

Comparing the number of children ever born of ‘non-consanguineous’ couples (i.e. F < 0.0156) to ‘consanguineous’ couples (i.e. F >= 0.0156), one finds opposite effects depending on land-ownership (Fig. 2): While number of children ever born is estimated significantly higher for ‘consanguineous’ couples in landowning families, estimates do not significantly differ (but seem to increase rather regarding parametric means) in the case of families being landless or with unknown amount of landholdings. However, consanguineous couples both among families with or without knonw landholdings do on average not have more grandchildren ever born (but rather fewer - although this estimated decrease is not shown to be significant).

Total fertility according to SES and consanguinity

Average number of births and grandchildren births. Note that estimates are highly skewed, i.e. mean (red boxes) differing from median.

Figure 2: Average number of births and grandchildren births. Note that estimates are highly skewed, i.e. mean (red boxes) differing from median.

Probability of grandchildlessness according to SES and consanguinity.

Figure 3: Probability of grandchildlessness according to SES and consanguinity.

Modeling grandchildlessness in a binomial model

The previously used sample has been restricted to those marriages contracted before 1830 in order to ensure that the birth of grandchildren can be sufficiently traced. The binary outcome of grandchildness is modeled in a mixed-intercept logistic regression allowing for baseline changes according to marriage periods and including consanguinity (non-consanguineous (F < 0.0156) vs. consanguineous (F >= 0.0156)) and landholdership (unknown/landless vs. landholding) as fixed effects interaction.

Extinction Risk Model
gc_extinct
(F > = 0.0156) 1.056**
(0.332)
with_grasenlandholding -0.573***
(0.130)
(F > = 0.0156):with_grasenlandholding -2.537*
(1.059)
Constant -1.855***
(0.102)
N 5,396
Log Likelihood -1,863.713
AIC 3,737.426
BIC 3,770.393
p < .05; p < .01; p < .001
Marginal effects of interaction terms in the mixed-effects logistic regression model.

Figure 4: Marginal effects of interaction terms in the mixed-effects logistic regression model.

Random effects for periods in the mixed-effects logistic regression model.

Figure 5: Random effects for periods in the mixed-effects logistic regression model.

Stillbirth rates and infant mortality

Stillbirth rates and infant mortality by SES and consangunity.

Figure 6: Stillbirth rates and infant mortality by SES and consangunity.

Breeding success and marriage probability of adult offspring

Breeding success and marriage Probability of offspring.

Figure 7: Breeding success and marriage Probability of offspring.

Probability of stillbirth and infant mortality in daughters

Probability of stillbirth and infant mortality in daughters from conasnguineous and non-consanguineous unions.

Figure 8: Probability of stillbirth and infant mortality in daughters from conasnguineous and non-consanguineous unions.

Probability of stillbirth and infant mortality in sons

Probability of stillbirth and infant mortality in sons from consanguineous and non-consanguineous unions.

Figure 9: Probability of stillbirth and infant mortality in sons from consanguineous and non-consanguineous unions.

Age at marriage of spouses and the protogenetic interval

Kaplan-Meier plots showing the proportion of brides and husbands still unmarried against their age and the protogenetic interval.

Figure 10: Kaplan-Meier plots showing the proportion of brides and husbands still unmarried against their age and the protogenetic interval.

Hypogamy risk

Johannes Johow
(Dr. rer. nat., Dipl.-Biol.)